Turning dataset into matrix by using mkmat

[Sungmin Cheu]

Dear all,
Hope all is well.

I am trying to turn values in the data set into matrix by using mkmat command.

However, as there are too many values in the varlist, it keeps failing to form a matrix.

Is there any way that I can drop some values under the varlist and form a matrix?

For example, Is there a way that I form a matrix with the top 10 values under a varlist?


Thanking you in advance.

[Andrew Musau]

Nothing stops you from using the commands preserve and restore to first restrict the number of observations, form the matrix and then revert to the original dataset.

For example, Is there a way that I form a matrix with the top 10 values under a varlist?

Do you mean top 10 observations or you are thinking of sorting the observations and picking the highest or lowest first? Remember that if you need to sort, you can sort in terms of only one variable in the varlist.

. sysuse auto
(1978 Automobile Data)

. preserve

. keep  price mpg weight length displacement


*SORT IN DESCENDING ORDER IN TERMS OF PRICE (HIGHEST TO LOWEST)
. gsort  -price mpg weight length displacement


. keep in 1/10
(64 observations deleted)

. mkmat price mpg weight length displacement, mat(auto)

. restore

. mat list auto

auto[10,5]
            price           mpg        weight        length  displacement
 r1         15906            21          4290           204           350
 r2         14500            14          3900           204           350
 r3         13594            12          4720           230           400
 r4         13466            14          3830           201           302
 r5         12990            14          3420           192           163
 r6         11995            17          3170           193           163
 r7         11497            12          4840           233           400
 r8         11385            14          4330           221           425
 r9         10372            16          3880           207           231
r10         10371            16          4030           206           350

If you are using an older version of Stata (I believe versions prior to Stata 12), you can increase the memory before starting your session if this is a space issue. See

help memory

[Sungmin Cheu]

Please refer to the below code or command for my question.
From the code, you will get a (2X10) matrix named c.
'in 1/10' indicates that you will form a matrix with values from the first row to 10th row under the varlist 1 and 2.

[code]:
mkmat varlist1 varlist2 in 1/10, matrix(c)



Oh! I did not know there was a reply to my question.
Thank you Mr.Musau for your reply! your code seems to work pretty good for me.

fyi, by top 10 observations, I just meant the observation from 1st row to 10th row.

[Andrew Musau]

Seems to be resolved now by your edit!

[Christophe Kolodziejczyk]

mkmat with the in qualifier is the shortest way, but here is another way

sysuse auto ,clear

local var price mpg weight length displacement
mata: st_matrix("c",st_data((1,10),"`var'"))
mat coln c = `var'
mat li c